Independent market research and competitive analysis of next-generation business and technology solutions for service providers and vendors Evolving to an Open C-RAN Architecture for 5G A Heavy Reading white paper produced for Fujitsu AUTHOR: STERLING PERRIN, PRINCIPAL ANALYST, HEAVY READING INTRODUCTION The next generation of mobile technology will place unprecedented demands on the efficiency, flexibility and scalability of the radio access network (RAN). These network demands – some of which are coming to light even today – are forcing a radical re-think of the entire RAN, including the remote radio head (RRH), the baseband unit (BBU) and the transport connec- tivity between the two. This white paper provides an assessment of the emerging C-RAN architecture with a focus on the practical evolutionary path that will take mobile operators from the distributed RANs of today to the fully-virtualized and open cloud RANs of the future. Key areas addressed are: • Drivers and benefits of a centralized BBU pool • C-RAN implications for fronthaul transport • Functional split options and the increasing importance of Ethernet fronthaul • Benefits and challenges of moving to a virtualized RAN • The need for an open RAN for virtualization BENEFITS OF A CENTRALIZED RAN ARCHITECTURE VS. DISTRIBUTED RAN Drivers for Radio Access Network Change Each new mobile technology generation has delivered greater data rates to mobile devices, and mobile users have moved more and more of their data usage from fixed devices to their smart phones, tablets and other devices. In addition to social media, video streaming is now common place on small screens. While the mobile data explosion has been a boon for the mobile industry, it is placing an unsustainable strain on the RAN. Operators must simultane- ously increase bandwidth to cell towers and mobile users while reducing cost for transport. The present mode of operation is the distributed RAN architecture, in which the RRH and the BBU RAN functions are all located at the cell sites and are all backhauled into a central switching center. This architecture was effective at 3G, but may not deliver the required combination of greater bandwidth at lower costs for advanced versions of 4G. At 5G, where data rates to devices will hit 1 Gbit/s, many mobile operators have concluded that the dis- tributed RAN architecture will not be viable. From Distributed RAN to Centralized RAN The C-RAN architecture was first proposed by China Mobile in 2009 as a better RAN architec- ture to handle the demands of high-bandwidth applications of the future. The "C" denotes both "centralized" and "cloud" – two separate but related aspects of this new architecture. Phase one – BBU centralization – is being done today by leading edge operators, including Verizon, AT&T, Telus, China Unicom, SK Telecom, KT, Japan's Docomo, among others. BBU centralization is a prerequisite for future BBU functional virtualization, but it also brings im- mediate benefits to operators, even with no virtualization involved. © HEAVY READING | SEPTEMBER 2017 | EVOLVING TO AN OPEN C-RAN ARCHITECTURE FOR 5G 2 Centralizing the BBUs provides several benefits to mobile network operators: • Economies of scale efficiencies: In the traditional RAN architecture, each cell site requires its own dedicated BBU, along with the associated power, cooling and routing functionality. Centralizing the BBUs in one location allows the BBU pool to share the same data center/central office physical space, batteries, electricity generators and cooling sources. Beyond the shared physical infrastructure, the BBU pool can be served by one large router, rather than separate, smaller routers required at each cell site. Instead of engineering each individual cell site for peak capacity, operators can engineer the BBU pool for peak capacity, allowing router ports to move from under- utilized BBUs in the pool to heavily-used BBUs as required by traffic demands. In short, the physical pooling ensures that all data center infrastructure and routing re- sources are used most efficiently and with the least amount of idle capacity and waste. • Reduced opex from sending technicians to cell towers for maintenance: Fur- ther operational savings from centralizing BBUs come from equipment and site maintenance. With a centralized RAN, technicians don't need to be dispatched out to cell towers for troubleshooting and routine maintenance of the BBUs. • Improved performance due to greater coordination among cells: Economies of scale and reduced maintenance are the most significant benefits from physical centralization, but centralization provides performance improvements, even without any virtualization. China Mobile, for example, has demonstrated up to a 30 percent performance increase from centralized BBUs using cell site aggregation. Performance gains include lower call drop rates, as well as increased downlink data rates and come from connecting RRHs in a base station of a cell to at least two base stations. Figure 1 illustrates the traditional distributed RAN architecture that places the RRH and the BBU at the cell tower and the new centralized RAN architecture that places only the RRH at the cell tower and moves the BBU functions to the centralized BBU hotel (or BBU pool). It also shows the efficiency gains of coordinated pooling compared to a distributed RAN. Figure 1: Distributed RAN & Centralized RAN Network Diagrams Source: Fujitsu & Heavy Reading © HEAVY READING | SEPTEMBER 2017 | EVOLVING TO AN OPEN C-RAN ARCHITECTURE FOR 5G 3 C-RAN IMPLICATIONS FOR FRONTHAUL TRANSPORT A C-RAN architecture consist of three main components: the baseband unit pool (BBU), the remote radio head (RRH), and the transport network connecting the RRH and the BBU, called the fronthaul network. In a distributed RAN, the RRH and BBUs are typically separated within the cell tower, with the RRH atop the tower and the BBU at the bottom. The Common Public Radio Interface (CPRI) protocol was standardized specifically for this connection. Physical Layer Options for Fronthaul While parts of the C-RAN architecture are still being defined, the physical-layer picture is clear: the C-RAN will be fiber-based. Still, within a fiber access network, there are several variations of dark fiber vs. wavelength division multiplexing (WDM) and active vs. passive components. Figure 2 highlights the various physical-layer options for C-RAN and some of the trade-offs of each. Figure 2: Trade-Offs of Different Physical Layer Approaches Physical Layer Description Pros Cons Dark Fiber Unlit fiber leased by the Great control of fiber Very inefficient use of fiber mobile operator, which assets. Works with any and capacity, particularly with then lights it up with its protocol or functional CPRI, as each site will require own optical equipment. split option. dozens of fibers and lacks any remote OA&M function. Passive WDM Colored pluggable SFPs Lower cost than active Requires pluggable colored (CWDM or DWDM) that WDM due to elimination optics in the BBU and RRH. don't require amplifiers of active components. CWDM not as scalable as between the BBU to More scalable than dark DWDM. Passive systems lack the RRH. fiber due to use of WDM. remote OA&M function. NG-PON2 ITU's GPON standard Lowest cost per bit of all Not as scalable as active WDM successor, allowing four C-RAN fiber options due option. Possible issues with wavelengths at 10 to reduction of active significant power loss through Gbit/s per wavelength components – reducing passive components. Lacks out to ONUs (symmet- both capex and opex. OA&M functions. NG-PON2 rical). Allows multiple PONs will allow reuse of products are not available. waves per tower/ONU the residential FTTH net- as needed. works. Low latency. Active WDM WDM transponders at Great efficiency with fiber Most expensive of the WDM each end of the 5G RRH capacity, great flexibility options due to number of and the BBU. in switching connections transponders and active com- and great scalability as ponents required at each end. new capacity required. Source: Heavy Reading, 2017 The migration from 4G radio to 5G (still being standardized) brings further implications and complications for the fronthaul network in C-RAN. Although it was designed only for short links, CPRI was initially expected to be the fronthaul protocol connecting RRHs and BBUs in C-RAN. However, vendors and operators quickly realized that CPRI, as it stands, does not © HEAVY READING | SEPTEMBER 2017 | EVOLVING TO AN OPEN C-RAN ARCHITECTURE FOR 5G 4 scale to the unique capacity and performance demands of 5G fronthaul. The primary front- haul challenges are detailed below. MIMO Scaling CPRI requires a dedicated link for every antenna – whether it's a dedicated fiber or, more likely, a dedicated wavelength on a fiber. This dedicated link requirement becomes problem- atic as radio vendors invest in multiple-input multiple-output (MIMO) technology that uses multiple transmitters and receivers to transfer data simultaneously, thus increasing data rates for 4G and 5G radio. To illustrate, a 2x2 MIMO transmission with three vectors requires 12 CPRI streams, as each input, output and vector requires its own CPRI link (i.e., 12 wavelengths). A 4x4 MIMO transmission doubles the count to 24 CPRI links/wavelengths. In fact, MIMO counts pro- posed for 5G will be significantly higher than these numbers. Initially, DWDM vendors loved the idea of CPRI links and WDM wavelengths everywhere, but it has become clear to the in- dustry that CPRI simply won't economically scale to the high antenna counts expected in 5G, due to MIMO. Functional Split Given the scalability challenge described above, the industry understands that a more effi- cient fronthaul technology is required for 5G. However, what that technology ultimately might be remains a matter of intense debate. A critical point that must be resolved before the technology moves forward is the "functional split" separating which Layer 1, 2, and 3 functions reside in the RRH and which of these processing functions reside at the BBU.